This past year we have continued to focus on a major part of long-term memory, termed semantic memory, that is composed of general information, such as facts, ideas, and the meaning of objects and words. We are particularly interested in characterizing the neural substrate mediating object and word meaning and its role in object perception. We are also interested in understanding how abstract knowledge, such as information about social interactions, is represented. Our studies have shown that information about salient properties of an object - such as what it looks like, how it moves, and how it is used - is stored in the perceptual and action systems active when that information was acquired. As a result, objects belonging to different categories such as animate entities (people, animals) and manmade manipulable objects (tools, utensils) are represented in partially distinct neural circuits. These distributed circuits also underpin our ability to understand more abstract events such as social and mechanical interactions. Although considerable progress has been made in understanding the neural circuits involved in representing concepts associated with concrete objects like animals and tools, little is known about how abstract concepts are represented in the brain. In collaboration with our colleagues at Emory University, we used fMRI to evaluate the neural representation of the abstract concepts 'convince' and 'arithmetic' using a newly developed concept-scene matching task that employed pictures of individuals in different situations to ground each concept in its typical context. Using this method we were able to show that brain regions known to be associated with understanding social interactions were active when subjects represented the abstract concept 'convince', whereas brain regions associated with understanding numbers were active when subjects represented 'arithmetic'. These results suggest that, like concrete entities, the meanings of abstract concepts arise from distributed neural systems that represent concept-specific content. In collaboration with our colleagues at Duke University, we have also engaged in a series of studies on the neural systems associated with acquisition of object-associated fear. Previous studies have documented changes in neural activity in the amygdala and sensory cortices when subjects learn to associate fear with a simple stimulus like a shape or color. However, real-world fears typically involve complex stimuli represented at the category level. As a result, an aversive experience with a particular object may lead one to infer that related instances, or exemplars, of that object category also pose a threat, despite variations in physical form (e.g., generalizing the fear associated with a specific dog, to all dogs). Using fMRI, we examined the effect of category-level representations of threat on human brain activity using the categories of animals and tools as conditioned stimuli. We found that activity in the amygdala and other category-responsive brain regions was modulated by the reinforcement contingency, leading to widespread fear of different exemplars from the reinforced category. Learning to fear animate objects was additionally characterized by enhanced functional coupling between the amygdala and fusiform gyrus. These findings provide novel evidence that aversive learning can modulate category-level representations of object concepts, thereby enabling individuals to express fear to a range of related stimuli. In marked contrast to a negative emotion like fear, other objects are typically associated with pleasurable experiences. A prime example is food. Food advertisements often promote choices that are driven by inferences about the pleasures of eating a particular food. Given the individual and public health consequences of obesity, it is critical to address unanswered questions about the specific neural systems underlying these pleasurable inferences. To address this issue, we used fMRI to measure neural activity when subjects made item-by-item ratings of how pleasant it would be to eat particular food depicted in a photograph. Consistent with previous studies, we found that activity in the orbitofrontal cortex was directly modulated by the subjects pleasantness ratings. In addition, we found that a subcortical brain region, the ventral pallidum, was also strongly modulated by the pleasantness ratings. This finding was particularly important because it demonstrated for the first time that a brain region known to be associated with reward in rodents, plays a central role in the moment-to-moment inferences of pleasure that influence our food-related preferences and decisions. In a related series of studies we evaluated the relationship between the visual system active when perceiving pictures of appetizing foods and the gustatory system active when tasting a pleasant liquid - apple juice. We found that many of the regions active when viewing food pictures, such as the anterior region of the insula cortex and orbitofrontal cortex, were also active when perceiving a pleasant taste. This finding was consistent with our perception/action model of how information about different object categories is stored and organized in the human brain. In addition, we found that primary gustatory cortex - the region of the brain where signals from the gustatory system first arrive in the cortex - was also active when subjects viewed pictures of food. Importantly, however, this response in primary gustatory cortex was dependent on the body's energy resources, as measured by levels of glucose in peripheral blood taken prior to the brain scanning session. In this way, signals from the body concerning energy needs may directly modulate the brains response to pictures of appetizing foods by heightened the response of the neural systems associated with reward and pleasure. Finally, in a series of studies we sought to take advantage of the ability of fMRI to record very slowly fluctuating neural signals when subjects are lying in the scanner without performing a specific task. Analysis of the data obtained during this task-free or resting state can reveal spatially distant regions of the brain showing similar patterns of fluctuating neural activity. The strength of similarity, or correlation, between this fluctuating activity in different brain regions can be used as a surrogate measure of the strength of the connections between them. We have applied this resting-state correlation analysis procedure to evaluate the strength of cross-hemispheric functional connectivity using fMRI data, and to explore similarities between task-free (rest) and task-dependent data recorded from sensory processing brain regions using magnetoencephalography (MEG), a recording technique that provides information on a much finer temporal time-scale (on the order of milliseconds) than provided by fMRI.